The invention provides a self-adaptive disturbing signal processing recognition device of a wide-domain all-fibre disturbance sensor network system, which is mainly used for the pipeline monitoring of safety detection and monitoring, etc. of pipelines delivering various risky or high-value chemical products. Based on an all-fibre reflection interference distribution type sensor circuit system, a method which adopts wavelet denoising and ICA signal separation and characteristic extraction is provided, with the realization process of two contents: on the one hand, in a data signal register, mixed signals carried with various noises and disturbance information are read out, the signal pre-processing and separation thereof are carried out, the characteristic value is then extracted, a KNN recognition module is entered finally, and the type and property of the input signal are recognized; on the other hand, when the signals to be recognized are processed and recognized, the signals are input into a recognized signal register, after the recognized result analysis of the signals to be input is completed, a main controller decides whether to carry out the manpower interception to the input signals so as to determine the type and property of the input signals further.